The interesting bit in the article for me was that if the account was required, and the lame-content posters were gone, the good-content posters also posted less (seemingly, because someone else wasn't wrong on the internet.)
However, while it is easy to measure, I'm not convinced that "number of good contributions" is a valid proxy for "quality of discussion" or "quality of content".
For example, some of those good contributions might have been repairing the damage made by bad ones. And on a less heated/combative forum, users may feel comfortable consolidating multiple small contributions into fewer but more well-thought out contributions. And perhaps even the good contributions that were prevented were (while good) on average lower quality than the ones that remained.